#

Key takeaways

• OpenAI analyzed more than 1.5 million work-related ChatGPT messages.

• The study covered messages from April through July 2026.

• Among roughly 6,200 consistently observed workers, recurring cross-occupation tasks increased over time.

• The share rose from 13.1% in April to 25.9% in July.

• OpenAI says the data was aggregated and anonymized.

What the study says

OpenAI published economic research on September 16, 2026. The study looks at work-related ChatGPT messages and focuses on how AI tasks change over time. It is presented as an early view of changing work patterns.

The research analyzed more than 1.5 million messages from April through July 2026. OpenAI also says it observed roughly 6,200 workers consistently during the period. The study reports that some cross-occupation tasks became recurring work.

The main finding

The study says previously used cross-occupation tasks grew as a share of occupation-specific AI activity. In April, they accounted for 13.1% of that activity. By July, the share had risen to 25.9%.

That change suggests some tasks were not one-off requests. Instead, they appeared more often across the observed period. The study also says recurrence varied by task.

What recurring work means here

In this report, recurring work refers to tasks that appear again after being used before. The study does not say every task followed the same pattern. It only reports that recurrence differed across tasks.

This matters because repeated tasks can signal a shift in how people use AI. A task may start as an occasional prompt and later become part of a regular workflow. That is the general pattern described by the research.

Data handling and limits

OpenAI says the messages were aggregated and anonymized. That means the study uses grouped data rather than individual message details. The report also frames the findings as an early view, not a final conclusion.

The source does not provide a forecast of employment outcomes. It also does not claim the findings predict specific labor changes. Readers should treat the results as descriptive research only.

Why the finding matters

The study points to a possible change in how AI is used at work. Some tasks may move from occasional support to repeated use. That can affect how teams think about workflow design and task planning.

The report does not give operational guidance. It does not name industries, job titles, or specific business functions. So the safest reading is general: repeated AI use may become more visible over time.

Morocco relevance

The source reports no Morocco-specific facts or employment outcomes. For readers in Morocco, the global lesson is to watch how recurring AI tasks may change daily workflows, while avoiding assumptions about local labor effects.

What to watch next

The study leaves several questions open. It does not explain why recurrence increased, or which tasks changed most. It also does not say whether the pattern will continue.

That means the report is best used as a snapshot. It shows a measurable shift in task repetition during a short period. It does not establish a long-term trend on its own.

Bottom line

OpenAI's research suggests that some cross-occupation AI tasks are becoming more recurring. The share of these tasks rose sharply across the study period. The findings are early, limited, and descriptive.

The strongest takeaway is simple. AI use at work may be moving from isolated prompts toward repeated task patterns. The study shows that shift in the data it reviewed, without making broader predictions.